Robust Machine Learning Systems: Challenges,Current Trends, Perspectives, and the Road Ahead

Abstract

Machine Learning (ML) techniques have been rapidly adopted by smart Cyber-Physical Systems (CPS) and Internet-of-Things (IoT) due to their powerful decision-making capabilities. However, they are vulnerable to various security and reliability threats, at both hardware and software levels, that compromise their accuracy. These threats get aggravated in emerging edge ML devices that have stringent constraints in terms of resources (e.g., compute, memory, power/energy), and that therefore cannot employ costly security and reliability measures. Security, reliability, and vulnerability mitigation techniques span from network security measures to hardware protection, with an increased interest towards formal verification of trained ML models.This paper summarizes the prominent vulnerabilities of modern ML systems, highlights successful defenses and mitigation techniques against these vulnerabilities, both at the cloud (i.e., during the ML training phase) and edge (i.e., during the ML inference stage), discusses the implications of a resourceconstrained design on the reliability and security of the system, identifies verification methodologies to ensure correct system behavior, and describes open research challenges for building secure and reliable ML systems at both the edge and the cloud.

Hardware and Software Optimizations for Accelerating Deep Neural Networks: Survey of Current Trends, Challenges, and the Road Ahead

“…Despite the great success and popularity of deep learning in recent years, recent researches showed that DNNs have intrinsic weaknesses that can threaten the security [267] [268] [269]. Starting from the work of Goodfellow et al [270], many researches have been conducted, with the purpose of identifying weaknesses (Adversarial Attacks) and their countermeasures (Adversarial Defenses) [271] [272].…”

Resistive Crossbar-Aware Neural Network Design and Optimization

“…Figure 18 (c) demonstrates the variations in the number of crossbars for different networks across cell bit-size (addressability). Increasing the cell addressability by one order of magnitude inversely reduces the number of crossbars by 93.72%, but the power and area costs associated with corresponding high-precision peripherals and analog-digital inter-conversion circuits 7 severely limit throughput improvements. In our design space exploration, w = 2 emerged as a sweet spot in terms of balancing the area and energy trade-offs for the test bench.…”

“…State-of-the-art networks rely on massive parameter count (≥ 10 8 ) and manually designed architectures (with 10 1 − 10 3 layers) to surpass prior benchmarks and beat human-level performance [2], [3]. However, these algorithms demand enormous memory (≥ 10 9 Byte), and operational cost (≥ 10 9 operations for each input), scaling commensurately with ever-growing datasets and network depth [4]- [7].…”

Machine Learning on the COVID-19 Pandemic, Human Mobility and Air Quality: A Review

“…Learning-based algorithms can retrieve meaningful features from a large volume of data to predict outcome accurately and are able to reveal the hidden patterns in the data set that were previously unknown [41]- [43]. At variance with traditional data processing systems, ML algorithms build models based on existing data with little or no distributional requirements for future predictions or decision making, which increases their performance tremendously [44].…”